Ksenia De Leo

• • Developing and integrating machine-learning models for clustering and classification on large, high dimensional datasets under high-noise conditions

• • Leading the development of a data-reconstruction algorithm that incorporates temporal information to combine fragmented data, reducing computational costs by ~30% while maintaining required performance

• • Conducting comprehensive data analysis and implementing deep learning models for anomaly detection, achieving 25–50% improvement in detection efficiency across several models

• • Developing and optimizing algorithms for noise reduction and signal separation in highly overlapping datasets, improving accuracy and robustness by up to 25%

• • Leading an international, cross-functional team of 15+ people across 5 countries to develop algorithms for next-generation timing detectors

• • Delivering complex analytical results in clear, actionable form to technical and non-technical audiences, including presentations at international conferences.

• • Designed and evaluated algorithms in C++/Python using digital twins to analyze system performance and support optimization of complex systems

• • Mentored and supervised 3 Master's and 5 Bachelor's students in data analysis and ML workflows, supporting successful project delivery.